INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Airflow obstruction is the hallmark of common chronic respiratory diseases including chronic obstructive pulmonary disease (COPD) and asthma (1). Spirometric testing is an essential component for diagnosing these conditions and it is increasingly being advocated for early detection of airflow obstruction, particularly among cigarette smokers (2, 3). The common occurrence of undiagnosed airflow obstruction in general practices and population-based surveys (4) suggests a need for widespread use of spirometry for early detection. However, little is known about the clinical relevance of undiagnosed airflow obstruction, and until this is defined along with the demonstration of therapeutic effectiveness, the role of detecting undiagnosed airflow obstruction with screening programs will remain controversial (8).

Many factors may be used to define the health impact of undiagnosed airflow obstruction, including physiological parameters, other measures of morbidity, and mortality. Based on opinion surveys and qualitative research, the concerns of people faced with chronic illnesses include maintaining the best possible functioning, minimizing symptoms, and caring for themselves (9). Therefore, to determine the relevance of undiagnosed airflow obstruction for patients, we used spirometry data, self-reports of respiratory symptoms, and health and functional status assessments that were obtained as part of the Third National Health and Nutrition Examination Survey (NHANES III) (10).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The detailed methods for NHANES III have been described previously (10, 11). Briefly, NHANES III was a cross-sectional, multistage probability sample representative of the total noninstitutionalized civilian population of the United States. The sample was selected from households in 81 counties during the period 1988 through 1994.

The NHANES III sample included 20,050 adults, 17 yr of age and older. The present analysis is restricted to 5,743 white participants from this sample (52% female), 45 yr of age and older (mean age 65.7 yr) who performed spirometry that met acceptability and reliability criteria of the American Thoracic Society (ATS) (see next section). Other racial and ethnic groups, including African Americans and Hispanics, were excluded because of sample size limitations.

Data Collection: Questionnaires and Spirometry

The NHANES III survey consisted of a household questionnaire and a health examination administered by trained examiners. Four categories of questionnaire items were used for this analysis including doctor diagnosis of chronic bronchitis, emphysema, and asthma; respiratory symptoms; cigarette smoking; and health impact (Appendix ).

Spirometry was performed with equipment that met ATS performance criteria (12). Procedures for testing were based on 1987 ATS recommendations (12), and included automated quality assessment during test performance. Spirometry was performed in 16,484 adults, and of these subjects, 1.7% had unusable curves (13).

Statistical Analysis

The NHANES III survey incorporates a complex probability sampling design in which certain segments of the population (e.g., blacks, elderly, very young) are oversampled to ensure statistically reliable estimates. Therefore, statistical analysis requires attention to information on the complex survey design and necessary weighting variables (11).

For this analysis, spirometry results were classified using modified ATS criteria (1). Obstructive airflow impairment was defined by an FEV₁/FVC ratio below the lower limit of normal (LLN) using equations from Hankinson and coworkers (13) and for individuals with airflow obstruction, the FEV₁ percent predicted value was also calculated using equations of Hankinson and coworkers (13) and severity of impairment classified using criteria modified from ATS (1) (Appendix ). The spirometry results were combined with questionnaire responses to further categorize subjects as no airflow obstruction, asthma, COPD, or undiagnosed airflow obstruction (Appendix ).

Data were weighted using the overall sampling weights, and point estimates were determined using SAS (14). The purpose of these analyses was to determine whether pulmonary function is an independent predictor of adverse health impact among subjects with undiagnosed airflow obstruction. Separate logistic regression analyses were conducted for each response in the health impact category. In each category, the dependent variable was the reporting of an adverse health impact with the following predictor variables: sex (male versus female); age ( 65 yr versus < 65 yr); FEV₁ (Stage Ia versus Ib, and II and III combined); comorbidities (cardiovascular disease, cancer, arthritis, diabetes); smoking status (ever versus never); and obesity (body mass index [BMI] greater than 27 kg/m² versus 27 kg/m² or less). The logistic regression analyses were conducted using SUDAAN (15).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Characteristics of Subjects

Overall, undiagnosed airflow obstruction was substantially more common than either COPD or asthma (Figure 1). Undiagnosed airflow obstruction was found in 12.0% of subjects compared with 3.1% with COPD and 2.7% with asthma. Undiagnosed airflow obstruction was more common among males (14.2%) than among females (9.9%) (Figure 1), but the prevalence was similar for all ages (data not shown).

View larger version (28K): [in this window] [in a new window]   Figure 1.   Distribution of white subjects, 45 yr of age and older, using self-reports of doctor-diagnosed COPD and asthma, and presence or absence of airflow obstruction based on spirometric results, NHANES III, 1988-1994.

Subjects with undiagnosed airflow obstruction had a higher prevalence of ever smoking (82.3%) than subjects with no airflow obstruction (54.2%) (Table 1). Furthermore, the prevalence of ever smoking was higher among males (89.5%) with undiagnosed airflow obstruction than among females (70.6%). The highest prevalence of ever smoking was found among subjects with COPD (93.9%), and was intermediate for subjects with asthma (76.5%).

Because of the potential impact of obesity on lung function and health status, the distribution of obesity was also examined in the different groups of subjects (Table 1). Undiagnosed airflow obstruction was associated with a lower prevalence of obesity (BMI greater than 27 kg/m²) than no airflow obstruction, 37.1% and 48.0%, respectively (Table 1). Similarly, obesity was less common among subjects with COPD (27.9%) and asthma (43.1%) than among subjects without airflow obstruction.

Health Impact of Undiagnosed Airflow Obstruction

Of subjects with undiagnosed airflow obstruction, 90.4% had mild impairment (Stages Ia and Ib), and 9.6% had moderate to severe impairment (Table 1). Except for dyspnea, the prevalence of respiratory symptoms increased among subjects with even the mildest impairment (Stage Ia) compared with subjects without airflow obstruction, and the occurrence of all symptoms consistently increased with increasing severity of FEV₁ impairment (Table 2). Dyspnea was the most common symptom at all stages of impariment, and was reported by over 50% of subjects with undiagnosed airflow obstruction when the FEV₁ fell below 75% of predicted.

Overall, among subjects with undiagnosed airflow obstruction, self-reports of adverse effects on health and functional status increased with severity of FEV₁ impairment (Figure 2A-2D). The reports of adverse impacts were higher than in subjects without airflow obstruction when the FEV₁ fell below 50% of predicted. At each stage of FEV₁ impairment below 75% predicted, subjects with COPD reported higher adverse health and functional effects compared with subjects with undiagnosed airflow obstruction.

View larger version (9K): [in this window] [in a new window]   View larger version (10K): [in this window] [in a new window]   View larger version (11K): [in this window] [in a new window]   View larger version (10K): [in this window] [in a new window]   Figure 2.   (A-D) Prevalence of adverse health effects among white subjects, 45 yr of age and older, for different stages of airflow obstruction, NHANES III, 1988-1994. The adverse health effects were (A) poor/fair general health status, (B) difficulty walking 1/4 mile, (C ) difficulty lifting or carrying 10 lb, (D) needing help with personal care. Solid square, no airflow obstruction; dashed line, undiagnosed airflow obstruction; solid line, COPD.

In logistic regression models (Table 3), with Stage Ia as the reference category and adjusting for sex, age, smoking, obesity, and other diseases, FEV₁ impairment below 50% of predicted was consistently and significantly associated with increased reports of impaired general health status and difficulty walking 1/4 mile. The association between stage of FEV₁ impairment and difficulty lifting and need for help with personal care was similar but not significant. The magnitude of the associations between stage of FEV₁ impairment and health and functional impact was similar or greater than those of being elderly, obese, or with associated diseases such as arthritis, cancer, cardiovascular disease, or diabetes.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this population-based sample of whites 45 yr of age and older, we found undiagnosed airflow obstruction to be more common than doctor-diagnosed COPD and asthma combined. After excluding subjects who reported a doctor-diagnosis of asthma or COPD, we identified 12% of the population as having airflow obstruction using spirometric criteria. Although the impairment was usually mild, approximately 5% of the 5,743 subjects had an FEV₁ less than 75% of predicted, which was associated with an increase in respiratory symptoms. Furthermore, after adjusting for smoking, obesity, and comorbid conditions, the risk of impaired health and functional status among subjects with undiagnosed airflow obstruction was independently associated with severity of FEV₁ impairment (Table 3). The occurrence of undiagnosed airflow obstruction with an FEV₁ less than 75% of predicted was similar among men (5.7%) and among women (4.6%). For males and females, ever smoking was strongly associated with all types of airflow obstruction, whether diagnosed or not. However, a substantial proportion of females with airflow obstruction (12.2% to 35.2%) never smoked, and 27.5% were undiagnosed (Table 1).

These estimates of the prevalence of undiagnosed airflow obstruction may be influenced by our criteria for defining airflow obstruction and the use of self-reports of doctor-diagnosed COPD and asthma. Restriction of the sample for this analysis limits the generalizability to whites, 45 yr of age and older. The goal of this restriction was primarily to identify subjects with undiagnosed COPD rather than those with asthma. Evidence to support this objective was provided by Dickinson and coworkers (7) who found 30 subjects with newly diagnosed airflow obstruction, 22 with COPD and eight with asthma, among 353 elderly patients aged 60 to 75 yr in the United Kingdom.

For this analysis, we used an LLN for the FEV₁/FVC ratio that declines with age (13), and although this is the method recommended by the ATS (16) for defining airflow obstruction few investigations have used this criteria (17). Often, a fixed FEV₁/FVC ratio of 70% is used, which results in an overestimation of the prevalence of airflow obstruction. For example, the overall prevalence of undiagnosed airflow obstruction using an FEV₁/FVC ratio of 70% or less in the NHANES III sample used for this analysis was 22.7% compared with 12% using an LLN that declines with age (Table 1).

Because of poor recall, self-reports of doctor-diagnosed COPD and asthma may have resulted in an overestimate of undiagnosed airflow obstruction, however, this misclassification is probably small. In a survey of patients with airflow obstruction conducted by van den Boom and coworkers (18), diagnostic misclassification from questionnaire responses was found in only 13% of cases. Further, doctor-diagnosed COPD misclassified as undiagnosed airflow obstruction would have little impact on the associations between undiagnosed airflow obstruction and health and functional status (Figure 2A-2D).

The prevalence of undiagnosed airflow obstruction has been described in several European countries (4), and the estimates have varied widely ranging from 3.2% to 11.8%. Differences in age groups, prevalence of smoking, and criteria for defining airflow obstruction may partly explain the variations in prevalence between these populations. Among a population-based sample of 1,196 Finnish subjects 65 yr of age and older, Isoaho and cowrkers (4) used an FEV₁/FVC ratio of 65% or less, and found an overall prevalence of undiagnosed airflow obstruction of 3.2%, 6.2% among men and 1.1% among women. Renwick and Connolly (5) conducted a survey of 783 patients 45 yr of age and older who were randomly selected from 22 general practices in the United Kingdom. Of these subjects 246 had spirometric testing and overall 11.8% had undiagnosed airflow obstruction (FEV₁/FVC ratio <65% for subjects less than 65 yr of age and < LLN for subjects 65 yr of age and older). In the Netherlands, van den Boom and coworkers (6) surveyed from general practices 1,155 subjects aged 25 to 70 yr without COPD, asthma, or other chronic diseases, and monitored them over a 6-mo period. Overall, 7.7% of this sample had undiagnosed airflow obstruction defined by FEV₁ levels at least two standard deviations below a predicted value on at least two occasions or bronchial hyperreactivity. Dickinson and coworkers (7) studied 353 subjects 60 to 75 yr of age from general practices in the United Kingdom, and using an FEV₁ level less than the lowest fifth percentile with less than a 9% increase in FEV₁ after a bronchodilator, found an overall prevalence of undiagnosed COPD of 6.2%.

In these earlier studies, limited information was available on the health and functional impact of undiagnosed airflow obstruction (4). Isoaho and coworkers (4) and Renwick and Connolly (5) examined respiratory symptoms in patients with chronic airways diseases, but neither separated subjects with a previous diagnosis and undiagnosed airways disease, and there were no other measures of health or functional status. In the Netherlands (18), 48 subjects with undiagnosed airflow obstruction (mean FEV₁ percent predicted 69.2%) had a lower disease-specific quality of life that was clinically relevant in the domain of fatigue, compared with subjects with respiratory symptoms but no spirometric impairment. Of the 22 newly diagnosed subjects with COPD identified by Dickinson and coworkers (7), FEV₁ percent predicted values were mild in 12 and moderate in 10, and scores on a respiratory impact questionnaire were normal in 16, mild in five, and moderate in one. These findings from previous studies, although limited, suggest that the severity of impairment may be greater in the U.S. sample. Further, the high occurrence of mild impairment (Stage Ia) and of asymptomatic subjects with more severe impairment in the NHANES III sample suggests that there is the potential for a large burden of illness that may manifest in the future.

There are many determinants of morbidity, including disease and social factors, and we were able to control for only some of these variables in our analysis (Table 3). Although we could not adjust for all of the factors in examining the impact of undiagnosed airflow obstruction, our findings suggest that self-reported impairment is associated with airflow obstruction that cannot be simply attributed to clustering of comorbidities in smokers.

While the relevance of self-reports of health and functional status may be questioned, reports of poor or fair health status, adjusting for age, socioeconomic status, and comorbidity, have been associated with a twofold increase in mortality (19). Moreover, mortality is increased with impaired lung function, (20), and obstructive lung diseases are a leading and growing cause of death in the United States (2, 25).

The greatest potential for lessening morbidity and mortality associated with undiagnosed airflow obstruction is likely to be in the areas of smoking cessation and antiinflammatory therapies. Although our analysis was limited to subjects 45 yr of age and older, available evidence suggests that the benefits of smoking cessation for prevention of impaired lung function are greatest at younger ages, usually before 45 yr of age (26- 29). In addition to improving lung function, which occurs within the first year of cessation (17), smoking cessation improves respiratory symptoms (30).

Asthma is associated with an accelerated decline of FEV₁ that may result in chronic airflow obstruction (31). This rapid decline in FEV₁ and the increased respiratory symptoms improve with inhaled corticosteroids (32). In addition to the underdiagnosis of cases of asthma in the general population that might benefit from early intervention, a substantial proportion of subjects with doctor-diagnosed asthma appears to be undertreated, based on our finding of a high prevalence of moderate to severe impairment of FEV₁, 25.0% of men and 37.7% of women (Table 1).

Rising mortality from COPD and results from the Lung Health Study (17) have prompted the development of a national strategy for the prevention, management, and research of COPD (2, 3). The Lung Health Study (17), a multicenter trial, demonstrated the efficacy of smoking cessation in slowing the rate of FEV₁ decline among current smokers with early COPD. The results of that study provide evidence for early detection with spirometry and intervention in current smokers, part of the national strategy for prevention of COPD (3). However, many unanswered questions about the use of spirometry for screening still remain (2). Our results of undiagnosed airflow obstruction in the NHANES III sample, for example, suggest that limitation of screening to current smokers will detect only a minority of the population with airflow obstruction, thus missing many opportunities to lessen morbidity and mortality from chronic airflow obstruction. The fact that over 80% of the subjects included in this analysis reported visiting a physician at least once during the previous 12 mo suggests that the clinical setting may provide a tremendous opportunity for a widespread program of early detection of airflow obstruction (3).